Three AI art controversies that change how you think about creativity
Ivy Wei | AI4Creativity Project
October 2018. Christie’s auction house, New York.
A blurry portrait of a man sells for $432,500. Where you’d normally expect an artist’s name, there’s a mathematical formula.
The work came from a Paris-based art collective called Obvious. Their announcement: this was the first AI-generated artwork ever sold at a major auction house.
Three days later, a 19-year-old programmer named Robbie Barrat posted on Twitter:
“Wait. They used my code.”
That tweet didn’t undo the sale. But it opened up a question that hasn’t gone away: when an algorithm is involved in making art, who actually made it?
Case One: Obvious
When creativity becomes a supply chain
What Obvious actually did was pretty simple. They took Barrat’s GAN code, which he had shared publicly on GitHub. They trained the model on portraits painted between the 15th and 20th centuries. They generated a large number of images, picked one, and sent it to auction.
Barrat’s objection was blunt: they hadn’t changed the code, they hadn’t retrained the model. They just ran the programme and chose an output. Does that count as creating something?
Obvious pushed back. The code was open source, they said. The real creative act was choosing the subject, curating the output, and getting the work taken seriously as art.
Break the process down and you get something interesting. This single artwork required:
- The developer who wrote the algorithm
- The historical artists whose paintings became training data
- The team who selected and framed the output
- The institution that gave it cultural legitimacy
Remove any one of those, and the work doesn’t exist. But if every part mattered, can authorship really belong to just one person or group?
What Obvious reveals isn’t an answer. It’s a structure. Here, creativity starts to look like a production line — one with shared labour, hidden contributors, and unequal credit.
Case Two: Jason Allen
When words become a craft
In 2022, a game designer named Jason Allen entered a piece called “Theatre D’Opéra Spatial” into the Colorado State Fair’s digital art competition. He made it using Midjourney. He won first place.
The backlash came fast.
“This isn’t art.” “He just typed a few words.” “The AI is the real artist.”
Allen’s response surprised a lot of people. He walked through his process in detail: dozens of hours exploring themes and styles, hundreds of iterations on his prompts, careful adjustments to composition and mood, and a long process of selecting from many generated images. The final prompt he used was over 200 words long, with technical parameters, style references, and narrative instructions.
We already accept photography, collage, and digital illustration as valid art forms. So here is the question: why can’t language be a visual medium too? Conceptual artists have been doing this for decades, but does AI prompting actually belong in that lineage?
But the deeper problem is whether the platform was quietly shaping his choices the whole time.
Allen’s case shifts the conversation. It’s no longer about whether a human was involved. It becomes: who is actually shaping the work’s direction over time?
Case Three: Botto
When the artist becomes a system
Botto calls itself a “decentralised autonomous artist.” The setup is straightforward enough: an AI generates hundreds of images each week. Community members vote using tokens. The image with the most votes gets minted as an NFT and auctioned. The results feed back into the AI, nudging its future outputs. By 2024, Botto had sold over a hundred works and made several million dollars.
Supporters say this is art becoming democratic — no single author, style emerging from collective choice.
Critics point out that collective voting tends to reward the safe, the familiar, and the predictable. Challenging or difficult work gets filtered out.
And “decentralised” doesn’t mean power has disappeared. Look more closely:
- The algorithm’s designers decide what the AI is even capable of making
- Voting influence depends on how many tokens you hold
- Early participants have shaped the aesthetic direction more than anyone who joined later
In Botto, the artist isn’t a person. It’s a set of rules. Power hasn’t gone away — it’s just been built into the system.
Three answers, all incomplete
Each case gives a different answer to the question of who the artist is:
- Obvious: the team that used the AI
- Allen: the individual who shaped the vision through prompts
- Botto: a hybrid of algorithm and community
Each of these works as an answer. None of them is fully satisfying. Because all three avoid the same underlying truth: making art has never been a solo act. It has always involved power, technology, institutions, and taste working together.
Renaissance painters relied on workshops and patrons. The Impressionists shaped each other through their social networks. Contemporary art stands on the entire history of the field and the institutions that support it. AI just makes this network visible, traceable, and impossible to ignore.
Maybe we’re asking the wrong question
Instead of asking “who is the artist?”, try these:
- What are the key nodes in this creative network?
- Whose contribution gets recognised, and who quietly disappears from the story?
- How is decision-making distributed across the system?
- Is the style being explored, or is it being trained into something predictable?
AI art hasn’t destroyed the idea of the artist. It’s just made it impossible to pretend the artist was ever a lone genius working in isolation.
Now it’s your turn
If you’ve made something with AI, the question probably isn’t “does this count as my work?”
It might be:
- Where do I sit in this chain?
- What decisions did I actually make?
- And which choices did I let the system make for me, without really noticing?
Your answer might look like Obvious, Allen, or Botto. Or it might be something that doesn’t have a name yet.
